About ReaxFF MD simulation errors

Dear Ray and Sanjib

I didn’t see anyone who considers non-reactive bond using ReaxFF MD.

I can think that is possible.

However, do you think that is impossible?

Your replys were very helpful for me.

Thanks for your replys.

Best Regards.

Dear Ray and Sanjib

I didn’t see anyone who considers non-reactive bond using ReaxFF MD.

I can think that is possible.

However, do you think that is impossible?

​the problem with defining bonds is, that - by default - LAMMPS will remove the pair of bonded atoms (and other pairs when bonds are connected) from the non-bonded neighbor list. however, since all bonds (and angles, dihedrals) in ReaxFF are implicit, this pruning of the neighbor lists, is resulting in incorrect computation of the ReaxFF forces part.

to recover the full non-bonded neighbor list, you have to change the settings for the bond topology based neighbor list exclusions with: special_bonds lj/cut 1.0 1.0 1.0

with this change LAMMPS will compute all regular forces as before, and then add the force from the explicitly defined bonds. so you cannot really have “non-reactive bonds​” using parameters from other force fields, but you can put a restraining force on such bonds. this alters the behavior of those bonds compared to those without the restraining force and thus the force parameters have to be very carefully chosen to keep that modification of the system to a minimum, yet at the same time achieve the desired restraining effect. you may want to consider using a customised bond potential, that is flat around the equilibrium distance and only yields a restraining force, when the bond is starting to break.

​axel.​

1 Like